The principal objective of this project is to elucidate the structure of the HIV-1 integrase protein, complexed with DNA and/or inhibitors, and to use the structural knowledge thus obtained to design better inhibitors of this enzyme with the goal of developing new anti-AIDS drugs. HIV integrase (IN) is the virally encoded enzyme responsible for integration of the retroviral DNA into the host genome. This step in the life cycle of HIV is essential for viral replication. Inhibition of integration is seen as an attractive target in the development of anti-AIDS therapies because no cellular homologue to IN is known, thus raising the hope that effective anti-IN based drugs with low-toxicity can be developed. The emergence of multidrug-resistant virus phenotypes during administration of cocktails of protease and reverse transcriptase (RT) inhibitors has further highlighted the need for alternative therapeutic approaches. IN is a 32kDa protein that is a product of the gag-pol fusion protein precursor contained in the virus particle. Upon completion of proviral DNA synthesis by RT, IN cleaves two nucleotides from each viral DNA end ("3'-processing"). After subsequent migration to the host cell's nucleus, IN catalyzes the insertion of the recessed 3'-terminus, generated during the 3'-processing step, into one strand of the host DNA. This reaction is termed 3' end joining (also known as integration or strand transfer) and occurs for both ends of the viral DNA simultaneously. The subsequent gap-joining is presumed to be performed by cellular DNA repair enzymes to yield a fully integrated proviral DNA. (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9083480&dopt=Abstract) Previous work, mainly based on 3D-pharmacophore searches in the NCI database, had yielded a number of inhibitors of IN. With the advent of more, and better, experimental structures (by X-ray crystallography and NMR) of HIV-1 IN as well as of closely related enzymes such as ASV integrase, it has become possible to model larger structures including multimeric models of the full-length protein, for which experimental structures are not available as of yet. We have generated such structures by means of molecular modeling techniques using all available experimental evidence. Special emphasis was placed on obtaining a model of the enzyme's active site with the viral DNA apposed to it as it might be after 3'-processing but before strand transfer, as described in (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=16075307&query_hl=1") Karki et al., 2004. This model is useful for structure-based inhibitor design of inhibitors which retain activity in vivo. We have made (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=12920196&dopt=Abstract) use of these structural models to study the potential binding modes of various diketo-acid HIV-1 IN inhibitors for which no experimental complexed structures are available. The results indicate that the diketo-acid IN inhibitors probably chelate the metal ion in the catalytic site and also prevents the exposure of the 3'-processed end of the viral DNA to human DNA. These models are currently being exploited for further inhibitor development using the resources described in our (http://ccrintra.cancer.gov/cms/annual_reports/printer_friendly_report.asp?ProjID=3876") database project, in particular through in silico screening of a database of more than 26 million purchasable screening samples.